Draggor | 9 Jul 22:02 2008
Picon

Installing cells-gtk3

I've got everything I need (I think), but on both windows and linux,
when I do (load "load") I run into this error:

debugger invoked on a ASDF:MISSING-COMPONENT in thread #<THREAD
"initial thread" {A72E739}>:
  component "gtk-ffi" not found

Type HELP for debugger help, or (SB-EXT:QUIT) to exit from SBCL.

restarts (invokable by number or by possibly-abbreviated name):
  0: [ABORT] Exit debugger, returning to top level.

(ASDF:FIND-SYSTEM :GTK-FFI T)

I have libcellsgtk, and there's the gtk-ffi folder in there too.  Help!

~Draggor
Draggor | 9 Jul 23:03 2008
Picon

Fixed old issue, onto some new ones

When I have everything loading in clisp on windows, I get this error:

*** - FIND-CLASS: GTK-BOOLEAN-TYPE does not name a class
The following restarts are available:
SKIP           :R1      skip 74 79 (DEFMETHOD
CFFI:TRANSLATE-TO-FOREIGN (VALUE #) ...)-6
STOP           :R2      stop loading file
C:\Programming\Lisp\cells-gtk3\gtk-ffi\gtk-ffi.fas
RETRY          :R3
Retry performing #<ASDF:LOAD-OP NIL #x19F33211> on
#<ASDF:CL-SOURCE-FILE "gtk-ffi" #x19F32685>.
ACCEPT         :R4
Continue, treating #<ASDF:LOAD-OP NIL #x19F33211> on
#<ASDF:CL-SOURCE-FILE "gtk-ffi" #x19F32685> as having been successful.
SKIP           :R5      skip (LET # # ...)
STOP           :R6      stop loading file
C:\Programming\Lisp\cells-gtk3\load.lisp
ABORT          :R7      ABORT

Under Linux and sbcl:

WARNING:
   COMPILE-FILE warned while performing #<COMPILE-OP NIL {B511171}> on
   #<CL-SOURCE-FILE "gtk-ffi" {B509CB1}>.

debugger invoked on a ASDF:COMPILE-FAILED in thread #<THREAD "initial
thread" {A72E739}>:
  erred while invoking #<COMPILE-OP NIL {B511171}> on
  #<CL-SOURCE-FILE "gtk-ffi" {B509CB1}>

(Continue reading)

Peter Hildebrandt | 12 Jul 23:59 2008
Picon

Re: Fixed old issue, onto some new ones

Hi Draggor,

sorry for the delay, Real Life keeps me busy these days.

Looks like a cffi version mismatch.  Did you pull cffi from darcs?  If
not, do so, delete all fasl files on your linux system, restart sbcl,
and try again.

If you have cffi from darcs, then there might be an issue with a
recent update in cffi itself which breaks something in gtk-ffi.  If
this is the case, I will have to investigate later -- currently I am
on my company laptop w/o linux and w/o lisp :-(

HTH,
Peter

On Wed, Jul 9, 2008 at 11:03 PM, Draggor <draggor <at> gmail.com> wrote:
> When I have everything loading in clisp on windows, I get this error:
>
> *** - FIND-CLASS: GTK-BOOLEAN-TYPE does not name a class
> The following restarts are available:
> SKIP           :R1      skip 74 79 (DEFMETHOD
> CFFI:TRANSLATE-TO-FOREIGN (VALUE #) ...)-6
> STOP           :R2      stop loading file
> C:\Programming\Lisp\cells-gtk3\gtk-ffi\gtk-ffi.fas
> RETRY          :R3
> Retry performing #<ASDF:LOAD-OP NIL #x19F33211> on
> #<ASDF:CL-SOURCE-FILE "gtk-ffi" #x19F32685>.
> ACCEPT         :R4
> Continue, treating #<ASDF:LOAD-OP NIL #x19F33211> on
(Continue reading)

Laziz Foo | 24 Jul 03:21 2008
Picon

installation notes, type error in test-gtk

I have installed cells-gtk3 following the very helpful instructions at:

http://www.washbear-network.de/peterblog/getting-started-with-cells-gtk/

and thought my notes might be useful.  I'm pretty much a noob, having
flirted on and off with lisp without ever getting serious.  For the
record, I'm running ubuntu 8.04, sbcl 1.0.11.debian.

0.  I followed the instructions at the address above.  (require
'test-gtk) gave me trouble on a foreign function call definition (I
think).  Decided to get the required packages from source control...

1.  CFFI from darcs requires alexandria, babel and trivial-features,
which I couldn't find via google, but noticed in the output from
cl-build:
"http://common-lisp.net/~loliveira/datcs/trivial-features".   Maybe
this is more obvious to others.

2.  libgtkglext1 is not installed by default  (fixed w/ apt-get)

3.  (start-app 'test-gtk) drops to the debugger with

The value of STRING is #(71 84 75 32 84 101 115 116 105 110
                         ...), which is not of type STRING.
   [Condition of type SIMPLE-TYPE-ERROR]

Backtrace:
  0: (SB-KERNEL:CHECK-TYPE-ERROR STRING #(71 84 75 32 84 101 ...) STRING NIL)
  1: (CFFI:FOREIGN-STRING-ALLOC #(71 84 75 32 84 101 ...))
  2: (GTK-FFI::GTK_WINDOW_SET_TITLE #.(SB-SYS:INT-SAP #X080C3A30) #(71
(Continue reading)

Ingo Bormuth | 24 Jul 09:43 2008
Picon

Re: installation notes, type error in test-gtk

> 1.  CFFI from darcs requires alexandria, babel and trivial-features,
> which I couldn't find via google, but noticed in the output from
> cl-build:
> "http://common-lisp.net/~loliveira/datcs/trivial-features".   Maybe
> this is more obvious to others.
>
> 3.  (start-app 'test-gtk) drops to the debugger with
> 
> The value of STRING is #(71 84 75 32 84 101 115 116 105 110
>                          ...), which is not of type STRING.
>    [Condition of type SIMPLE-TYPE-ERROR]

IIRC, this is due to changes in cffi (from 0.9.2 to darcs). 
See my mail from June, 23rd on the issue.

The particular patch is called 0005-Upgrade-CFFI-to-current-darcs-0.9.2.patch
it is also attached to this mail.

Ingo

_______________________________________________________________________
Jetzt neu! Sch├╝tzen Sie Ihren PC mit McAfee und WEB.DE. 30 Tage
kostenlos testen. http://www.pc-sicherheit.web.de/startseite/?mc=022220

> 1.  CFFI from darcs requires alexandria, babel and trivial-features,
> which I couldn't find via google, but noticed in the output from
> cl-build:
(Continue reading)

Laziz Foo | 25 Jul 04:44 2008
Picon

Re: installation notes, type error in test-gtk

Thanks for the patch-- that worked!  One other thing was required:
changing the gtk-object 'container' slot's accessor to 'container-of'
in widgets.lisp per your email in June.  The demo runs perfectly after
that change.

-Laziz
Kenny Tilton | 25 Jul 08:36 2008
Picon

Re: [cells-gtk-devel] installation notes, type error in test-gtk

Laziz Foo wrote:
>> While you are waiting for a genuine GTk person to respond, all I can say is
>> that it looks as if the code in CFFI that automatically tries to do the
>> right thing with Lisp data to get it into a form the C function will
>> tolerate is failing to do the right thing. Since this code is pretty
>> common/well tested, all I can think is that perhaps a recompile in and
>> around the specific function gtk-window-set-title (gwst) might do the trick.
>> Or just bail on your Lisp, restart, then do a full rebuild from scratch
>> (ASDF :force t) IIRC. Possibly during your brave struggles (congrats on
>> making it this far, btw) you got one thing compiled before something else
>> and the order should have been the opposite. pardon the hand-waving.
>>
>> I presume you are running demo code untouched, or I would worry that "GTK
>> Testing" should be wrapped in with-c-string or something.
>>
>> Congrats again and thanks for the notes.
>>
>> kt
> 
> 
> With the patch supplied by Ingo, and the widgets.lisp change mentioned
> in his June email, I'm good to go.  Thanks for the kind words, and
> your entertaining blog entries  <at>  smuglispweenie.

Thanks! So I have lost track, is this Cells-Gtk3 (with Cells 3 Inside) 
or Oldskool Cells-Gtk?

Cells3 integrity clearly is to be preferred, but I miss the good old 
days of Cells Classic when I just SETfed here and SETFed there and let 
the data converge /eventually/ on correct values. :)
(Continue reading)

Kenny Tilton | 25 Jul 17:38 2008
Picon

Re: [cells-gtk-devel] installation notes, type error in test-gtk

Laziz Foo wrote:
>> Thanks! So I have lost track, is this Cells-Gtk3 (with Cells 3 Inside) or
>> Oldskool Cells-Gtk?
> 
> 
> This is cells-gtk3.  Oldschool works out of the box, but I wanted the
> cairo stuff.

And you got help from someone other than Peter H so it sounds as if CG3 
has taken hold. Cool.

So it is a battle to death between Cello and Cells-Gtk3 for GUI 
supremacy? Awesome.

>  It's a case of scratching my own itch--  every
> stock/futures charting app I've used (both windows and linux) are
> inadequate in ways that are important to me; might as well write my
> own.
> 
> Your propaganda has convinced me that somehow I would have
> greenspunned Cells,  or given up on the project entirely while chasing
> state issues around.

Ah, good, the spam is working. Yes, the paradigm is inevitable so one 
ends up automating dataflow one way or another. Over on c.l.l someone is 
explaining how Qt's signals/slots eliminate the need for Cells. 
Whatever. I just like the mounting evidence that dataflow is inevitable. 
  I would add it to the prior art in the Cells manifesto but it is 
getting ridiculous, I might just edit that down to "It's everywhere."

(Continue reading)

Ingo Bormuth | 30 Jul 00:30 2008
Picon

Problem with sbcl-1.0.18_[Scanned]

Hi,

sbcl-1.0.18 fails with

  #.(SB-SYS:INT-SAP #X00000000) fell through ETYPECASE expression.
  Wanted one of (CHARACTER SYMBOL INTEGER).

whenever +C-NULL+ is used.

This is triggered for example by line 463 in widgets.lisp:
(gtk-window-set-icon-from-file (id self) new-value +c-null+)))

Sbcl-1.0.17 works fine. Sbcl-git is also broken. I use cffi-darcs btw.

Does that ring a bell for anybody? Does anybody already use 1.0.18?

Thanks.

--

-- 
Ingo Bormuth,  voicebox & telefax:  +49-(0)-12125-10226517
PGP public key 86326EC9 at http://ibormuth.efil.de/contact


Gmane